Space-based Observations of the ice Sheets: Ice mass changes evaluated from 2013 to present. – SOSIce

We use the ESA’s Sentinel-1 radar (every 6 days), Sentinel-2 optical (every 5 days) and USGS’s Landsat-8 optical (every 16 days) acquisitions to determine surface ice speed on a weekly basis. This work will provide a precise record of the time evolution of ice motion and should allow us to learn more about the seasonal and long-term signals that models need to consider, the external forcings at different time-scale and, of course, the ice discharge evolution from sub-annual to decadal time-scale.
We will use elevation measurements from CryoSat-2/ESA, ArcticDEM, TanDEM-X/DLR, ATM/NASA and IceSAT-2/NASA to establish the evolution of the surface elevation on a monthly to annual basis.
Developing realistic ensemble simulations with the state-of-the-art ice flow model Elmer/Ice, developing metrics to score the model outputs and assimilating the new observations obtained by the project.
Assimilation of ice free surface and ice speed time series using ensemble data assimilation methods, based on the ensemble Kalman filter.

We first created a time series on a land-terminating sector in Greenland at high temporal resolution. We then studied different post-processing methodologies in order to generate calibrated and simplified time series. We show in a publication by Derkacheva et al. that it is possible to obtain time series with a temporal resolution of 2 weeks and improve the accuracy by a factor of three over the original measurements. We have established annual time series allowing to establish the long-term evolution of Greenland glaciers. These time series have been included in 2 recent studies of Greenland glacier evolution (Khan et al. 2020; Wood et al. 2021), both of which suggest that current projections may underestimate Greenland's contribution to sea level rise.

We evaluated the potential of using inverse methods on a dense time series of ice speed to assess the seasonal evolution of basal conditions. Our results show that these inversions can be used to understand the seasonal evolution of basal conditions (Derkacheva et al. submitted). We also focus on the development and evaluation of ensemble methods to better quantify the effect of uncertainties in projections. These uncertainties come from poorly-constrained physical processes or model initializations. We gathered a set of satellite data for a case study on Upernavik Isstrøm and for comparison with our ensemble simulation. This study showed that the average of our simulation ensemble had similar performance to other published studies. However, the advantage of using an ensemble approach compared to these works is that our results are not deterministic. We will continue to work on the physical processes that would allow the ensemble to better reproduce the glacier retreat. Then we will evaluate more precisely which parameters have a significant influence. This work will allow us to see if it is possible to implement assimilation methods to further reduce the uncertainty in order to obtain more reliable projections.

By taking advantage of the continuous observations and by assimilating them in an ice flow model, we will follow the ice sheet evolution in a fundamentally new way compared to current approaches. Significant technical and scientific issues would be solved from the results of this project, including securing the capacity to process large quantities of data for ice sheet studies, better understanding of the underlying physical processes causing increased in glacier ice discharge, improving ice-sheet model initialization before computing projections, and precisely reassessing the sea-level budget.

1. Derkacheva, A.; Mouginot, J.; Millan, R.; Maier, N.; Gillet-Chaulet, F. (2020) Data Reduction Using Statistical and Regression Approaches for Ice Velocity Derived by Landsat-8, Sentinel-1 and Sentinel-2. Remote Sens., 12, 1935. doi.org/10.3390/rs12121935
2. S.A. Khan, A.A. Bjørk, J.L. Bamber,, M. Morlighem, M. Bevis, K. H. Kjær, J. Mouginot, A. Løkkegaard, D. M. Holland, A. Aschwanden, B. Zhang, V. Helm, N.J. Korsgaard, William Colgan, N.K. Larsen, Lin Liu, K. Hansen, V. Barletta, T.S. Dahl-Jensen, A.S. Søndergaard, B.M. Csatho, I. Sasgen, J. Box & T. Schenk (2020) Centennial response of Greenland’s three largest outlet glaciers. Nat Commun 11, 5718. doi.org/10.1038/s41467-020-19580-5
3. M. Wood, E. Rignot, I. Fenty, L. An, A.A. Bjørk, M. van den Broeke, C. Cai, E. Kane, D. Menemenlis, R. Millan, M. Morlighem, J. Mouginot, B. Noël, B. Scheuchl, I. Velicogna, J.K. Willis, H. Zhang (2021). Ocean forcing drives glacier retreat in Greenland. Science Advances. doi.org/10.1126/sciadv.aba7282
4. A. Derkacheva, F. Gillet-Chaulet, J. Mouginot, E. Jager, N. Maier & S. Cook (submitted) Seasonal evolution of basal environment conditions of Russell sector, West Greenland, inverted from satellite observation of surface flow, The Cryopsh. Disc. doi.org/10.5194/tc-2021-170

Sea level rise has become a challenge facing society due particularly to the ice sheets that are contributing more significantly than previously anticipated. The mass loss of the ice sheets is due to increased surface melt runoff and outflow of ice associated with current climate warming. Here, we propose to improve our understanding of the dynamic component. Indeed, the ice discharge modulated by changes in ice velocity and thickness changes remains the largest uncertainty in the current and future contribution of the ice sheets to sea level rise. Quantifying and understanding the past/present/future contribution of the ice sheets to sea level rise under the current warming climate requires answering fundamental questions as: How has the ice velocity, thickness and so discharge of outlet glaciers changed on sub-annual to decadal time scales? What are the main and most important external forcing that are controlling changes in glacier ice discharge into the ocean? How can we use ice dynamic observations of the recent past to teach numerical ice flow model and get more precise projection of sea level rise?
Until recently the answers to those questions were limited, mainly because the ice sheet observations were spatially incomplete and temporarily sparse, resulting in averaged products to maximize spatial coverage at the expense of temporal information. However, in the last few years, we entered a new era of spaceborne ice sheet observations with the launch of the ESA’s CryoSat-2 in 2010, USGS’ Landsat-8 in 2013 and ESA’s four Sentinel-1 & 2 between 2014 & 2016. Used in the synergistic manner, these satellites offer the first chance for sustained, continuous data acquisition over the ice sheets to map ice motion and elevation.
Taking the opportunity offered by these new satellites, the SOSIce project will reconstruct at high temporal and spatial resolution the ice flow for the largest glaciers of Greenland and Antarctica to refine mass balance estimates and improve the forecasting skills of the numerical ice flow models. We have envisioned this work in 3 successive steps: derive time series of the (1) dynamical and geometrical structure of the glaciers from these new sensors, (2) assimilate them into the state-of-the-art ice flow model Elmer/Ice, and (3) disseminate our results using public data archive for the scientific community.
By taking advantage of the continuous observations and by assimilating them in an model, we will follow the ice sheet evolution in a fundamentally new way compared to current approaches. Significant technical and scientific issues would be solved from the results of this project, including securing the capacity to process large quantities of data for ice sheet studies, better understanding of the underlying physical processes causing increased in glacier ice discharge, improving ice-sheet model initialization before computing projections, and precisely reassessing the sea-level budget. This project will set very good grounds to initiate an international, scientific collaborative effort to facilitate the growth and establishment of the novel and rapidly growing field of remote sensing of the cryosphere over large datasets.